
In the age of AI reasoning, training smarter, more capable models is critical to scaling intelligence. Delivering the massive performance to meet this new age requires breakthroughs across GPUs, CPUs, NICs, scale-up and scale-out networking, system architectures, and mountains of software and algorithms.
In MLPerf Training v5.1 - the latest round in a long-running series of industry-standard tests of AI training performance - NVIDIA swept all seven tests, delivering the fastest time to train across large language models (LLMs), image generation, recommender systems, computer vision and graph neural networks.
NVIDIA was also the only platform to submit results on every test, underscoring the rich programmability of NVIDIA GPUs, and the maturity and versatility of its CUDA software stack.
NVIDIA Blackwell Ultra Doubles Down The GB300 NVL72 rack-scale system, powered by the NVIDIA Blackwell Ultra GPU architecture, made its debut in MLPerf Training this round, following a record-setting showing in the most recent MLPerf Inference round.
Compared with the prior-generation Hopper architecture, the Blackwell Ultra-based GB300 NVL72 delivered more than 4x the Llama 3.1 405B pretraining and nearly 5x the Llama 2 70B LoRA fine-tuning performance using the same number of GPUs.
These gains were fueled by Blackwell Ultra's architectural improvements - including new Tensor Cores that offer 15 petaflops of NVFP4 AI compute, twice the attention-layer compute and 279GB of HBM3e memory - as well as new training methods that tapped into the architecture's enormous NVFP4 compute performance.
Connecting multiple GB300 NVL72 systems, the NVIDIA Quantum-X800 InfiniBand platform - the industry's first end-to-end 800 Gb/s networking platform - also made its MLPerf debut, doubling scale-out networking bandwidth compared with the prior generation.
Performance Unlocked: NVFP4 Accelerates LLM Training Key to the outstanding results this round was performing calculations using NVFP4 precision - a first in the history of MLPerf Training.
One way to increase compute performance is to build an architecture capable of performing computations on data represented with fewer bits, and then to perform those calculations at a faster rate. However, lower precision means less information is available in each calculation. This means using low-precision calculations in the training process calls for careful design decisions to keep results accurate.
NVIDIA teams innovated at every layer of the stack to adopt FP4 precision for LLM training. The NVIDIA Blackwell GPU can perform FP4 calculations - including the NVIDIA-designed NVFP4 format as well as other FP4 variants - at double the rate of FP8. Blackwell Ultra boosts that to 3x, enabling the GPUs to deliver substantially greater AI compute performance.
NVIDIA is the only platform to date that has submitted MLPerf Training results with calculations performed using FP4 precision while meeting the benchmark's strict accuracy requirements.
NVIDIA Blackwell Scales to New Heights NVIDIA set a new Llama 3.1 405B time-to-train record of just 10 minutes, powered by more than 5,000 Blackwell GPUs working together efficiently. This entry was 2.7x faster than the best Blackwell-based result submitted in the prior round, resulting from efficient scaling to more than twice the number of GPUs, as well as the use of NVFP4 precision to dramatically increase the effective performance of each Blackwell GPU.
To illustrate the performance increase per GPU, NVIDIA submitted results this round using 2,560 Blackwell GPUs, achieving a time to train of 18.79 minutes - 45% faster than the submission last round using 2,496 GPUs.
New Benchmarks, New Records NVIDIA also set performance records on the two new benchmarks added this round: Llama 3.1 8B and FLUX.1.
Llama 3.1 8B - a compact yet highly capable LLM - replaced the long-running BERT-large model, adding a modern, smaller LLM to the benchmark suite. NVIDIA submitted results with up to 512 Blackwell Ultra GPUs, setting the bar at 5.2 minutes to train.
In addition, FLUX.1 - a state-of-the-art image generation model - replaced Stable Diffusion v2, with only the NVIDIA platform submitting results on the benchmark. NVIDIA submitted results using 1,152 Blackwell GPUs, setting a record time to train of 12.5 minutes.
NVIDIA continued to hold records on the existing graph neural network, object detection and recommender system tests.
A Broad and Deep Partner Ecosystem The NVIDIA ecosystem participated extensively this round, with compelling submissions from 15 organizations including ASUS, Dell Technologies, Giga Computing, Hewlett Packard Enterprise, Krai, Lambda, Lenovo, Nebius, Quanta Cloud Technology, Supermicro, University of Florida, Verda (formerly DataCrunch) and Wiwynn.
NVIDIA is innovating at a one-year rhythm, driving significant and rapid performance increases across pretraining, post-training and inference - paving the way to new levels of intelligence and accelerating AI adoption.
See more NVIDIA performance data on the Data Center Deep Learning Product Performance Hub and Performance Explorer pages.
North America Stories
05/12/2025
NEW YORK Iris, the new cloud-connected camera control platform, has officially launched with features that turn virtually any PTZ camera into a software-connect...
05/12/2025
WASHINGTON The Federal Communications Commission has approved AT&T's $1.02 billion acquisition of spectrum from UScellular in a decision that was issued sho...
05/12/2025
The Best Coldplay Songs: 21 Tracks That Shoot for the Stars From Yellow to Viva La Vida, Fix You to Paradise, this playlist goes back to the start.
December ...
05/12/2025
Zafris Lecture Series Brings Nabil Ayers to Berklee The 32nd annual James G. Zafris Distinguished Lecture series was held on Thursday, November 13 with guest ...
04/12/2025
SVG Sit-Down: ProximaVision's Claudio Lisman on Why Tethered Drones Could Be...
04/12/2025
SVG Campus Shot Callers: Imry Halevi, Senior Associate Director of Athletics, Co...
04/12/2025
Platinum White Paper: LiveU Lightweight Sports Production: A Step Change in Spor...
04/12/2025
London to Riyadh: DAZN brings the boxing glamour to new production levels for Be...
04/12/2025
Analysis: Paramount bets on the battering ram' with Champions League play By Callum McCarthy, Editor-at-Large
Tuesday, December 2, 2025 - 10:12
Print ...
04/12/2025
Space City Home Network Launches SCHN DTC App for Astros and RocketsThe Rockets and Astros were previously the lone NBA and MLB teams without a DTC appBy Jason...
04/12/2025
SVG Summit 2025 Preview: Content Workflows Workshop Spotlights Evolution of Spor...
04/12/2025
New Sponsor Spotlight: Geotech's Patrick Wambold On the Unreal Engine Revolu...
04/12/2025
Curt Gowdy Jr. - Master Storyteller, Nationally and RegionallyBy Jason Dachman, Editorial Director, U.S.
Thursday, December 4, 2025 - 1:52 pm
Print This Sto...
04/12/2025
(L-R) Rebecca Lichtenfeld, Mohammadreza Eyni, Sara Khaki, and Judith Helfand att...
04/12/2025
Coronal mass ejections caused by eruptions on the surface of the sun can have fa...
04/12/2025
Gracenote Content Connect enables media ecosystem to precisely align ad campaigns and programming based on rich content signals
NEW YORK - December 4, 2025 - N...
04/12/2025
Lightware, a global specialist in AV connectivity, is looking back on a year defined by new advancements, strong collaboration and continued growth. Across the ...
04/12/2025
Riedel Communications today announced a new partnership with Haivision, a leading global provider of mission-critical, real-time video networking and visual col...
04/12/2025
Harmonic (NASDAQ: HLIT) and Normann Engineering today announced a major milestone in their strategic collaboration, celebrating 20 successful broadband deployme...
04/12/2025
Creative software developer Foundry today announced Mari 7.5, the latest iteration of its artist-friendly paint toolset that can handle large, detailed assets w...
04/12/2025
Professional Wireless Systems (PWS), a leading provider of wireless audio solutions and RF management, was on site at Dreamforce 2025 in San Francisco providing...
04/12/2025
LTN's purpose-built IP video network brings all-movie diginet to over 100 stations and streaming platforms in just three months while eliminating satellite ...
04/12/2025
Bitmovin, the leading provider of video streaming solutions, today announced a strategic partnership with ThinkAnalytics, the global leader in AI-powered data a...
04/12/2025
The HELM, a global expert in cinematic live broadcast and high-end production workflows, has signed a partnership agreement with Keslow Camera, one of North Ame...
04/12/2025
At ISE 2026, LiveU will showcase its expanded IP-video EcoSystem, enabling broadcasters, sports, production companies and pro-AV professionals to share their st...
04/12/2025
Since the beginning of commercial television, advertising has been a key part of broadcasting. Over the years, the technology for inserting ads into programs ha...
04/12/2025
MUNICH and MILAN Warner Bros. Discovery said HBO Max is expanding into Germany, Italy, Austria, Switzerland, Luxembourg and Liechtenstein on Jan. 13, 2026, and ...
04/12/2025
SAN FRANCISCO AudioShake has launched its first streaming-capable software development kits (SDKs) designed specifically for real-time music detection and copyr...
04/12/2025
NASHVILLE The mobile and REMI production company TNDV has announced that it headed south into Mexico to live-produce the three-day 2025 Zane Grey Championship P...
04/12/2025
BURBANK, Calif. Hollywood Professionals Association Executive Director Phil Kubel has stepped down from the organization to pursue new opportunities, the group ...
04/12/2025
WASHINGTON The Federal Communications Commission said it has closed 2,048 inactive proceedings, the largest number of dormant dockets ever terminated in a singl...
04/12/2025
A new tech blog from Netflix highlights the importance of the AV1 open video codec, which now powers about 30% of the platform's streaming and discusses a v...
04/12/2025
Back to All News
Hugo Silva, Leonor Watling, Esther Acebo and Gorka Otxoa Star ...
04/12/2025
Back to All News
Step Inside the World of Troll 2: VFX Breakdown Featuring Dire...
04/12/2025
Developers, researchers, hobbyists and students can take a byte out of holiday s...
04/12/2025
Editor's note: The Game Pass edition of Hogwarts Legacy' will also be supported on GeForce NOW when the Steam and Epic Games Store versions launch on t...
04/12/2025
Scientists find cancer weak spot in backup DNA repair system New findings from Scripps Research reveal how certain tumors survive DNA damage and point to a stra...
03/12/2025
MLS Cup 2025 Production To Feature Four iPhone 17 Pros as Game-Coverage CamerasStay tuned to SVG on Friday for our in-depth story on this year's MLS Cup pro...
03/12/2025
SVG LIVE! 2025: All Sessions Now Available to Watch on SVG PLAYThe inaugural event placed a spotlight on the exciting world of live entertainmentBy SVG Staff
...
03/12/2025
(L-R) Peter Scriver and Seth Scriver introduce their documentary Endless Cookie for its premiere at the Egyptian Theatre in Park City. (Photo by Andrew H. Wa...
03/12/2025
COLUMBIA, Md. Lionsgate and its TV syndicator subsidiary Debmar-Mercury have selected LTN to launch and deliver the new MovieSphereGold all-movie digital networ...
03/12/2025
VIENNA, Austria Video streaming solutions provider Bitmovin and ThinkAnalytics, a provider of AI-powered data analytics for TV, have formed a strategic partners...
03/12/2025
SAN FRANCISCO & THE COLONY, Texas Dolby Laboratories is making what it is calling a new chapter in its retail efforts as part of an agreement with NFM (Nebras...
03/12/2025
MONTREAL Grass Valley has delivered a 4K Ultra-High-Definition (UHD) outside broadcast (OB) truck to Guangdong Radio and Television (GRT), in partnership with B...
03/12/2025
Our users spoke and we listened. December's Maxon One release delivers long-awaited improvements across Cinema 4D, Redshift, ZBrush, and Red Giant.
Whether...
03/12/2025
DHD audio reports a successful 2025 with new additions to its range of broadcast-quality audio production, post-production and routing equipment.
DHD innovatio...
03/12/2025
Grass Valley recently delivered a cutting-edge 4K Ultra-High-Definition (UHD) outside broadcast (OB) truck to Guangdong Radio and Television (GRT), in partnersh...
03/12/2025
CINCINNATI GatesAir has introduced Maxiva XTK, an update to its Maxiva XTE software-defined TV exciter. XTK is a new cost-efficient model primarily developed fo...
03/12/2025
WUPPERTAL, Germany Riedel Communications is partnering with Haivision, a global provider of mission-critical, real-time video networking and visual collaboratio...
03/12/2025
WASHINGTON The Federal Communications Commission has opened a docket for comments on the proposed $6.2 billion Nexstar acquisition for Tegna and set deadlines f...